I noticed that the stats for the discriminatory phrase count were being parsed/calculated from the actual listing descriptions each time a query was run from /tools/stats. The discriminatory phrase count variable was being set by calling listing.check_phrase for each phrase rather than relying on the established discriminatory phrase_listings table. This seems redundant to me, is slow, and does not produce the same results as a join between the listings and phrases table (through phrase_listings). I believe the correct approach is to rely on data in the existing table, otherwise I'm not sure what the function of the phrase_listings table is.
I may be a bit confused, but it also appears that bin/rake classify_listings:initial_set sets discriminatory:true if listing.illegal? is true..but listing.illegal only returns true if listing.discriminatory
is true. It doesn't appear to actually classify listings as discriminatory?
I noticed that the stats for the discriminatory phrase count were being parsed/calculated from the actual listing descriptions each time a query was run from /tools/stats. The discriminatory phrase count variable was being set by calling listing.check_phrase for each phrase rather than relying on the established discriminatory phrase_listings table. This seems redundant to me, is slow, and does not produce the same results as a join between the listings and phrases table (through phrase_listings). I believe the correct approach is to rely on data in the existing table, otherwise I'm not sure what the function of the phrase_listings table is.
I may be a bit confused, but it also appears that bin/rake classify_listings:initial_set sets discriminatory:true if listing.illegal? is true..but listing.illegal only returns true if listing.discriminatory is true. It doesn't appear to actually classify listings as discriminatory?